On 02.02.2008 02:46, Jeremy Wright wrote:
Yes, the server gets a DHCPREQUEST response to its DHCPOFFER. I took the ethereal advice and found that not only is DHCP working, but TFTP is working as well. I noticed it was loading the nbi.img image (sucessfully), so I
Good. How big is nbi.img?
switched it to pxelinux.0 and got a different response. The client starts to load the pxelinux.0 image, gets to block 2, and then resets, asks for an address via DHCP again, and starts trying to download the image again until it gets to block 2. It will keep going in this loop forever unless I stop it. When it loads the nbi.img file, it doesn't have the looping problem.
That looks a lot like memory problems, but it is only a gut feeling. (One image is loaded completely , another image with a possibly different load address causes a crash.) A size comparison and load address comparison could help debug the issue.
I'm a little puzzled. Should I try another version of Etherboot? I'm using version 5.2.6.
There is a small chance that the behaviour of Etherboot 5.4.3 is different. Please try it.
Is there any chance you can try to have etherboot load etherboot? The problem with that is of course that the second etherboot has to send a different ID string for its DHCP requests, otherwise you can't find out whether the new DHCP request is from the flashed etherboot after the machine resets itself or from the etherboot loaded over the network.
However, my experience with etherboot is very limited, but Ron and others know a LOT about it.
Regards, Carl-Daniel